Page 376 - Discrimination at Work The Psychological and Organizational Bases

14. LAW AND PSYCHOLOGY
343
bias) and because relevant neighborhoods may be predominantly White.
These causes may even evince an unconscious prejudice against members
of a protected class, but because intent is irrelevant to a showing of dis­
parate impact, such issues are irrelevant in disparate impact cases. The
disparate impact model therefore does not provide a way for unconscious
intent to be considered in employment discrimination, either as a plaintiff
sword or as an employer shield. It is concerned only with the consequences
of certain employment practices on protected classes and seeks to elimi­
nate "artificial, arbitrary, and unnecessary barriers to employment when
the barriers operate invidiously to discriminate" (Griggs v. Duke PowerCo.,
1971, p. 431).
The disparate impact model acts in mysterious ways, as can easily be
seen by examining some statistical issues regarding it. A nonobvious char­
acteristic of the disparate impact model is that it does not require proof that
an employer's screening mechanism has an actual impact on the workplace
(nor is absence of actual impact an employer defense (Connecticut v. Teal,
1982)). Most organizations rely on the use of multiple screening mecha­
nisms, particularly for hiring decisions. Because there can be redundancy
in the use of multiple screens, no one screen may be the actual cause of the
impact. The plaintiffs only have to prove that the screen would have had
a disproportionate impact on the members of the plaintiff's class if it were
applied independently of any other screens used in the selection process in
order to establish their prima facie case, however. For example, if word-of-
mouth recruiting would have screened the same individuals as a required
test, then the use of the word-of-mouth recruiting would have been redun­
dant and the disparate impact could have been "caused" by the test. The
net effect of the recruiting method would be zero. However, a challenge
to the recruiting method could lead to its removal if it could not be busi­
ness justified. Thus, the disparate impact model provides the somewhat
counterintuitive result that a screening method can be viewed as having an
adverse impact even though it has no actual (net) impact on any members
of the protected class in the employer's work force. The employer's lia­
bility in the form of make-whole remedy (e.g., backpay), however, would
extend only to those members of the workforce who experienced an ac­
tual effect, which in this case would be no one (Paetzold & Willborn,
1996).
The situation can become even more complicated by a further examina­
tion of the redundancy of screens. For example, it is easy to demonstrate
that an employment screen can have a disparate impact on a protected
group even though it has the net actual effect of benefiting the group.
Consider the example in Table 14.1. The word-of-mouth recruiting ap­
proach yields a selection rate of .49 for Black persons and .70 for White